ClawHub
Back to skill

Security audit

Cloud Infra Automation

Security checks across malware telemetry and agentic risk

Overview

This skill is for cloud infrastructure automation, but it includes broad auto-trigger terms and an unguarded production destroy example that users should review carefully before installing.

Review this skill before installing. Use it only where an agent is allowed to run infrastructure commands, and require explicit approval and review steps before any apply or destroy operation, especially against production cloud accounts.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)

Vague Triggers

Medium
Confidence
95% confidence
Finding
The trigger keywords are very broad and generic terms such as 'cloud', 'azure', 'automation', and 'management', which are likely to match ordinary user conversations unrelated to this specific skill. In an agent platform, this can cause unintentional invocation of a skill that has exec capability, increasing the chance of inappropriate command suggestions or execution in unrelated contexts.

Missing User Warnings

High
Confidence
97% confidence
Finding
The quick-start examples include a destructive command ('./cloud.sh destroy prod') without any warning, confirmation requirement, or explanation of the consequences. In the context of infrastructure automation and exec-enabled skills, this is dangerous because users or agents may treat the example as safe boilerplate and accidentally delete production resources, causing outages and irreversible data loss.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.